AI Tech Lead
**关于 Lucidya** Lucidya 为 Agentic 客户体验构建了操作系统。我们的旗舰产品线是一个代理平台 (Agent Studio),客户在阿拉伯语和英语中建立并运行 AI 代理，以解决文字和语音频道中的真实客户体驗问题。**角色摘要** AI 技术领导是负责我们代理平台的 AI 层质量，可靠性和技术方向的高级工程师。你将设置一个拥有区域的 AI 工程师团队的技术杆，并且你自己会编写生产代码。你需要戴2个帽子，其中一个是 AI 帽子，另一个是领导力**关键责任** 技术方向 - 对人工智能架构和利用能力负责：模型选择，提示模式，检索设计，护和语音管道。- 用明确的理由做出并记录技术决定。质量，评估和可靠性 - 定义评估标准并从实验到生产转移关键工作 - 引入 Agentic 开发的领先测试框架 - 在客户报告之前确保生产监控和在线评价发现质量下降，幻觉和延迟峰值。团队领导和工程实践 - 领导 AI 团队设计，开发和提供我们的 Agentic 平台 - 导师人工智能工程师，通过配对和反来提高技术标准，并为招聘做出贡献。实践交付 - 主要调试涉及人工智能的关键生产事件的原型/实验建设以验证新功能 （新模型，代理模式，语音特性）在团队承诺之前。跨职业合作伙伴关系 - 与产品一起工作将客户使用情况转化为技术要求 - 与工程协作在共享平台的工作：工具调用，工作流程，部署和扩展。思维和行为 - 拥有者：从端到端采取结果，包括团队之间的差距，并且不被问过。- 竞争判断：根据企业需求做出决定，并可以解释质量，速度和成本之间的竞争。- 建设性挑战：舒适地质疑现有方法并提出更好的方法。- 团队参与者：分享信誉，帮助同事取得成功，将团队目标放在个人胜利之前。- 创造力与创新：通过"外盒"思维来解决复杂的挑战，成为创新的冠军。
查看英文原文
**About Lucidya**
Lucidya builds an operating system for Agentic customer experience. Our flagship product line is an agentic platform (Agent Studio) where customers build and run AI agents that resolve real customer experience issues across text and voice channels, in Arabic and English.
**Role summary**
The AI Tech Lead is the senior hands-on engineer accountable for the quality, reliability and technical direction of the AI layer of our agentic platform. You will set the technical bar for a team of AI engineers who each own an area, and you will write production code yourself. You will need to wear 2 hats, one is the AI hat and the other is leadership
**Key responsibilities**
Technical direction
- Own and be accountable for the AI architecture and harness: model selection, prompting patterns, retrieval design, guardrails and voice pipeline.
- Make and document technical decisions with clear rationale.
Quality, evaluation and reliability
- Define the evaluation standard and move key work from experimentation to production
- Introduce leading testing frameworks for Agentic development
- Ensure production monitoring and online evals catch quality drops, hallucinations and latency spikes before customers report them.
Team leadership and engineering practices
- Lead the AI team in designing, developing, and delivering our Agentic Platform
- Mentor AI engineers, raise the technical bar through pairing and feedback, and contribute to hiring.
Hands-on delivery
- Lead debugging of critical production incidents involving AI behaviour.
- Manage the building of prototypes/experiments to validate new capabilities (new models, agent patterns, voice features) before the team commits to them.
Cross-functional partnership
- Work with Product to turn customer use cases into technical requirements
- Work with Engineering on shared platform work: tool calling, workflows, deployment and scaling.
Mindset and Behaviors
- Ownership: takes outcomes end to end, including gaps between teams, and follows through without being asked.
- Trade-off judgement: grounds decisions in business needs and can explain the trade-offs between quality, speed and cost.
- Constructive challenge: comfortable questioning existing approaches and proposing better ones.
- Team player: shares credit, helps teammates succeed, and puts team goals ahead of individual wins.
- Creativity and innovation: solve complex challenges with 'out-of-box' thinking and be a champion for innovation
- Experience: 7+ years in software or ML engineering, including 2+ years building LLM-based applications in production. Not just building AI point solutions, AI platform level experience is required.
- Technical leadership: has led a team of 3+ AI engineers/scientist technically (design reviews, standards, mentoring)
- LLM application engineering: deep hands-on experience with prompting, structured outputs, guardrails, agent patterns/harness development (planning, tool use, multi-step reasoning).
- Ability to work in ambiguous settings: can bring clarity and structure to complex situations
- Evaluation: has built LLM evaluation pipelines (golden datasets, LLM-as-judge, regression suites) and used them to gate releases.
- Production engineering: strong Python; experience with APIs, cloud deployment, observability and debugging distributed systems.
- Communication: fluent written and spoken English; able to write clear design docs and decision records.
Nice-to-have qualifications
- Arabic NLP: experience with Arabic and its dialects, especially Gulf Arabic, in LLM or speech systems.
- Voice AI: experience with ASR, TTS and real-time voice agents, including latency optimisation.
- Production monitoring for LLMs: hands-on use of tracing and online eval tools (for example Langfuse, Arize or similar).
- Model adaptation: fine-tuning or distilling open-weight models, and serving them cost-effectively.